Obtaining Social Relationship Type of Network Subjects

1. A method for alleviating a need of mutual operation during a process of determining a social relationship type among network subjects, the method comprising: obtaining network social behavior data of the network subjects whose social relationship type is to be determined; discretizing a respective network social behavior type into multiple levels; calculating a respective weight of the respective network social behavior type based on a ratio of total network subjects contacted by a respective network subject to a number of contacts with which the respective network subject has the respective network social behavior type; calculating a respective conditional probability that a respective level of the respective network social behavior type corresponds to a respective social relationship type; calculating a respective merged probability of the respective social relationship type between the network subjects according to the respective conditional probability; and automatically determining a social relationship type based, at least in part, on: a largest merged probability as an initial estimate of the social relationship type between the network subjects, and the respective weight of the respective network social behavior type.

2. The method as recited in claim 1 , wherein the calculating the respective conditional probability comprises: obtaining data including existing network social behavior data of existing network subjects whose social relationships are known as an initial sample data set; discretizing the respective network social behavior type in the existing network social behavior data into the multiple levels according to a quantity range of the respective network social behavior type in the existing network social behavior data; and calculating the respective conditional probability that the respective level of the respective network social behavior type corresponds to the respective social relationship type according to the data in the initial sample data set.

3. The method as recited in claim 2 , wherein the quantity range of the respective network social behavior type is a range of a number of occurrence times of the respective network social behavior type.

4. The method as recited in claim 1 , further comprising determining the social relationship type with the largest merged probability as a final estimate of the social relationship type between the network subjects in response to determining that the largest merged probability is larger than a preset threshold value.

5. The method as claimed in claim 4 , further comprising determining the preset threshold value, the determining including: obtaining data including existing network social behavior data of existing network subjects whose social relationships are known as an initial sample data set; obtaining a required accuracy; calculating a merged probability of respective network subjects in the initial sample data set; choosing a particular value as a hypothetical threshold value; estimating a social relationship type of the respective network subjects; and comparing an estimated result with a real social relationship in the initial sample data set to obtain an accuracy of the estimated result to determine the preset threshold value.

6. The method as recited in claim 5 , further comprising using the particular value as the preset threshold value in response to determining that the accuracy of the estimated result is equal to or substantially equal to the required accuracy.

7. The method as recited in claim 5 , further comprising decreasing the particular value in response to determining that the accuracy of the estimated result is larger than the required accuracy.

8. The method as recited in claim 5 , further comprising increasing the particular value in response to determining that the accuracy of the estimated result is smaller than the required accuracy.

9. The method as recited in claim 1 , further comprising calculating a social relationship strength index of a second network subject relative to a first network subject in the respective social relationship type according to the respective weight of the respective network social behavior type and a respective frequency of the respective network social behavior type.

10. The method as recited in claim 9 , wherein calculating the respective weight of the respective network social behavior type by using a formula 〈 social ⁢ ⁢ network ⁢ ⁢ behavior ⁢ ⁢ weight 〉 = log ⁢ total ⁢ ⁢ network ⁢ ⁢ subjects ⁢ ⁢ contacted ⁢ ⁢ by ⁢ ⁢ a ∑ x = B X ⁢ [ if ⁢ ⁢ 〈 a , x 〉 ⁢ ⁢ is ⁢ ⁢ involv ⁢ ⁢ ed ⁢ ⁢ in ⁢ ⁢ k , then ⁢ ⁢ ‘ 1 ’ ⁢ ⁢ else ⁢ ⁢ ‘ 0 ’ ] wherein: a represents the first network subject; b represents the second network subject; and k represents the respective network social behavior type.

11. The method as recited in claim 9 , further comprising calculating the respective frequency of the respective network social behavior type by using a formula 〈 network ⁢ ⁢ social ⁢ ⁢ behavior ⁢ ⁢ frequency 〉 = logNum ⁢ 〈 a , b 〉 ⁡ [ k ] log ⁢ ∑ x = b x ⁢ ∑ k = 0 n ⁢ Num ⁢ 〈 a , x 〉 ⁡ [ k ] wherein: a represents the first network subject; b represents the second network subject; n represents a total number of social behavior types between the first network subject and the second network subject; k represents the respective network social behavior type; Num<a, b>[k] represents a number of occurrence times of the respective social behavior K between the first network subject and the second network subject; and an antilogarithm portion of a denominator represents a total number of occurrence times between the first network subject and all other network subjects.

12. The method as recited in claim 9 , comprising classifying the respective social relationship into a sub-type according to one or more factors including a data range of the social relationship strength index of the second network subject relative to the first network subject in the respective social relationship.

13. An apparatus for alleviating a need of mutual operation during a process of determining a social relationship type among network subjects, the apparatus comprising: an obtaining unit that obtains network social behavior data of the network subjects whose social relationship type is to be determined; a discretizing unit that discretizes a respective network social behavior type into multiple levels; a calculating unit that calculates a respective conditional probability that a respective level of the respective network social behavior type corresponds to a respective social relationship type and calculates a respective merged probability of the respective social relationship type between the network subjects according to the respective conditional probability; a strength calculating unit that calculates a respective weight of the respective network social behavior type based on a ratio of a total network subjects contacted by a respective network subject to a number of contacts with which the respective network subject has the respective network social behavior type; and an initial estimating unit that automatically determines a social relationship type based, at least in part, on: a largest merged probability as an initial estimate of the social relationship type between the network subjects, and the respective weight of the respective network social behavior type.

14. The apparatus as recited in claim 13 , further comprising: an initial sample data set obtaining unit that obtains data including existing network social behavior data of existing network subjects whose social relationships are known as an initial sample data set; a level dividing unit that discretizes the respective network social behavior type in the existing network social behavior data into the multiple levels according to a quantity range of the respective network social behavior type in the existing network social behavior data; and a conditional probability calculating unit that calculates the respective conditional probability that the respective level of the respective network social behavior type corresponds to the respective social relationship type according to the data in the initial sample data set.

15. The apparatus as recited in 13 , wherein the quantity range of the respective network social behavior type is a range of a number of occurrence times of the respective network social behavior type.

16. The apparatus as recited in claim 13 , further comprising a final estimating unit that determines the social relationship type with the largest merged probability as a final estimate of the social relationship type between the network subjects in response to determining that the largest merged probability is larger than a preset threshold value.

17. The apparatus as claimed in claim 16 , further comprising a preset threshold value determining unit that: obtains a required accuracy for an initial sample data set; calculates a merged probability of respective network subjects in the initial sample data set; chooses a particular value as a hypothetical threshold value; estimates a social relationship type of the respective network subjects; compares an estimated result with a real social relationship in the initial sample data set to obtain an accuracy of the estimated result to determine the preset threshold value; and using the particular value as the preset threshold value in response to determining that the accuracy of the estimated result is equal to or substantially equal to the required accuracy.

18. The apparatus as recited in claim 13 , wherein the strength calculating unit is further configured to calculate a social relationship strength index of a second network subject relative to a first network subject in the respective social relationship type according to the respective weight of the respective network social behavior type and a respective frequency of the respective network social behavior type.

19. The apparatus as recited in claim 18 , further comprising a social relationship classifying unit that classifies the respective social relationship into a sub-type according to one or more factors including a data range of the social relationship strength index of the second network subject relative to the first network subject in the respective social relationship.

20. One or more memories stored thereon computer-executable instructions executable by one or more processors to perform operations for alleviating a need of mutual operation during a process of determining a social relationship type among network subjects, the operations comprising: obtaining network social behavior data of the network subjects whose social relationship type is to be determined; discretizing a respective network social behavior type into multiple levels; calculating a respective weight of the respective network social behavior type based on a ratio of total network subjects contacted by a respective network subject to a number of contacts with which the respective network subject has the respective network social behavior type; calculating a respective conditional probability that a respective level of the respective network social behavior type corresponds to a respective social relationship type; calculating a respective merged probability of the respective social relationship type between the network subjects according to the respective conditional probability; determining a social relationship type with a largest merged probability as an initial estimate of a social relationship type between the network subjects; and automatically determining the social relationship type based, at least in part, on: the largest merged probability as a final estimate of the social relationship type between the network subjects in response to determining that the largest merged probability is larger than a preset threshold value, and the respective weight of the respective network social behavior type.